verne.freenode.net changed the topic of #mlpack to: http://www.mlpack.org/ -- We don't respond instantly... but we will respond. Give it a few minutes. Or hours. -- Channel logs: http://www.mlpack.org/irc/
zoq|sherlock is now known as zoq
travis-ci has joined #mlpack
< travis-ci>
mlpack/mlpack#158 (master - ddb8ae9 : Ryan Curtin): The build passed.
< naywhayare>
the funny thing is, the netbook I am using right now is actually so slow that it's quicker to check in my code and see what travis does with it than to compile it locally...
< naywhayare>
I should probably just be more patient though...
< zoq>
I thought you build everything on a remote machine
travis-ci has joined #mlpack
< travis-ci>
mlpack/mlpack#162 (master - baeb388 : Ryan Curtin): The build passed.
< naywhayare>
usually, but today I am traveling (I'm at ICML in Lille) so internet is bad
< zoq>
ah right, the cf framework presentation
< naywhayare>
yeah... trying to get some comparisons with other libraries before the presentation on friday
< naywhayare>
I'd like to write some benchmarking suite scripts for them, but for now what I'm doing is pretty ad-hoc
< zoq>
If you like you can send me the parameter set and then I can write the scripts, but I'm not sure I get it done before the presentation.
< naywhayare>
sure, if you like, but it's a little bit complex in this case
< naywhayare>
the graph I wanted to generate was RMSE over time, so I had to use MaxIterationTermination with more and more iterations, and look at the resulting RMSE
< naywhayare>
it's more than just a number, it's a time series, so I'm not sure how that could be encoded in the benchmark system
< naywhayare>
on the other hand, a good general benchmark might just be RMSE and time with default parameters
< naywhayare>
when I have better internet tomorrow at the conference I'll send you some more details of what I'm doing, if you wnat
< zoq>
that would be nice, I guess we can write a RMSE metric module like Anand did and run the module to produce some results over time
< naywhayare>
it'll be easier than that since the cf program produces a line that says "RMSE" when you give it a test set :)